Currently the routine for multiplication of unsigned chars in the Z80 port extends the input to 16 bits and calls the 16 bit multiplication routine.
The attached patch contains an 8-bit multiplication routine. It is faster than the current one. The patch will speed up 32 bit integer multiplications, too since the 32 bit multiplication routine calls the 8-bit routine.
I have run the regression tests and found no problems. The patch is against 2.7.5 #5019.
Log in to post a comment.